In this presentation, Austen Heinz explains Cambrian Genomics' workflow for producing massive numbers of custom DNA sequences at very low cost. He begins by describing how microarray platforms synthesize trillions of DNA molecules in parallel, noting that only about 30% of the strands are correct while the rest are erroneous, creating a mixed “soup.” To isolate the correct sequences, the mixture is partitioned onto individual beads, each ideally carrying a single strand, and amplified. The beads are then fixed onto a glass slide where a color‑coded sequencing‑by‑color method reads each base by adding and cleaving fluorescently labeled nucleotides, generating a positional map for every bead. Using the identified coordinates, a laser‑pulse catapulting system ejects only the beads that contain verified sequences. The retrieved short fragments (≈100 bp) are then assembled into longer DNA constructs through either bulk bead pooling or robotic well‑by‑well assembly, followed by cleanup steps. Heinz emphasizes that the chemistry is decades old and unpatented, making the raw reagents cheap, and that the entire pipeline—synthesis, quality control, retrieval, and assembly—enables scalable, low‑cost DNA production for custom orders.
Chapters
So does anyone have any interest in learning how this actually works, like how we make DNA for free? Sure. Okay. I'll try not to get too technical, but we're making millions and millions of different unique sequences at a time. We're actually making trillions of molecules on a machine called a microarray. There's different ways to, to make lots and lots of DNA, different processes. Some use electrical chips like we use, others use light-based methods, others use inkjet printers to build these things, but they're stacking letters on top of each other to make chains. And that's cool, that technology's been around since about 1994 or so, you know, launched right here. probably right down the street in Silicon Valley, because some guy from the microfabrication industry realized that you could stack— I think that means my mouse is— or my clicker is going to stop working soon. But some guy from the silicon manufacturing industry realized that making DNA layer by layer was very similar to the way you make transistors layer by layer. So, And so anyway, this mass production of DNA has existed for a while. The problem is, is that most of the DNAs made using these massively parallel processes for DNA manufacturing are wrong. So in our array, approximately 30% are correct per run. 30% of the individual strands are correct. 70% are wrong. On top of that, They're all mixed up together, trillions of molecules, millions of uniques that we tried to make. And so we got this messy, dirty soup. And so we have to kind of get rid of this complexity. But to take a step back and make this even more simple, think about how anything today is manufactured. How is it made? Have you seen the show How's It Made? It's pretty much the same episode Every time. It's a hot dog, it's a sock, and it's the same, but it's the same process every time. So how is anything made? Some kind of assembly line where you do one piece, you pass that piece on, another thing adds to it. Yeah, you have assembly line, but basically anything that's made at an industrial scale is made either massively parallel or huge quantities of, right? But do they just ship them out the door after they throw that together? No. There's always, in whatever one of these things that you see, there is a vision system that's looking at these things and assembly workers plucking out— pulling off the bad ones and leaving the good ones, right? Yeah. Doing quality control, right? And so anything that you make, whether it's socks, hot dogs, golf clubs, who knows, it gets made at scale, thrown on a line, you look at the good ones, the bad ones, and you select the good ones, and the bad ones go to the incinerator, right? That's the same with everything. It's the exact same with this process. You can't make anything at scale without doing quality control. So after we make this DNA on array, which by the way is basically free, the chemistry to do it is about 40 years old, so there's no patents. The bottles, the A, T, C's and G's, the bottles that those chemicals come in are actually more expensive than the letters inside, right? It's dirt cheap. Then we go to our next technology, which is what's good and what's bad. So we use a process called DNA sequencing. Right, to be able to read the letters. Does that make sense? Anyone have any questions so far? Okay. Would you just be reading one at a time that way? No, we're reading many at a time. So, after we read, we make all this DNA, you got trillions of soup of trillions of molecules with only 30% or so being correct, all mixed up together. So, what we do is we separate this soup So, instead of making a soup, we separate the strands onto beads. So, each bead gets one strand. Some beads get no strands. So, and some beads get 2 or 3 strands, which isn't good, by the way. And then we copy— so, but we aim to get one piece of DNA on one bead. Afterwards, we make millions of copies of the DNA on the bead. And then we stick it to a piece of glass. Now we've got about a billion beads, about a billion sequences on glass. Next thing, we read these. And the way that we read these is we have a color letter— color letters that we use to build the second strand. So we've attached one strand of DNA, we've made many copies, And those copies, they're double-stranded. We basically remove that second strand from the bead, and then we build the second strand using color letters. So color A, color T, color C, color G. So if we're looking, say, at position 1, if there's a red— if this bead is red and this bead is green, that means there's an A here and there's a T there. And this is, you know, yellow, and that's orange, that's C, and that's G. And that's looking at one position. Then you clip that color off, and then you add to the second position. So DNA, A matches with T, G matches with C, right? So by knowing the color that's in position 1, we know what's here. By knowing the color position here, we know what this is, and so we know what this is. Does that make sense? Yeah. You're basically reading it by adding colors that you're clipping off after you add each letter. Any questions about that? So after we do that, we know what everything is. So this bead right here has a unique ID marker. It has a Cartesian, has an XY on a piece of glass. So over this, just like bingo, basically. It even kind of looks like bingo now because we use patterned arrays. So it could be like A34, B19, except there's a billion of them, so that system doesn't really work too well. So after we know what everything is, well, it's time to collect the good ones. The bad ones, who cares? They're going to get thrown away. They just remain on the plate, and then after we're done, we just throw that away. To get these sequences off, we use a process called laser pulse catapulting, which creates an explosion on the surface of glass, releasing the sequences that we want, the ones that are correct, that we defined on the computer. So when someone sends us an email, it's got a long file of all these different genes that they want. Those get— they're about maybe 1,000 or 1,500 letters— get broken down into 100-base-pair pieces. By our just computer software, and then we make those 100-base pair sequences on our array. We take those molecules attached to a billion beads, then we read those 100-base pair molecules on beads. Once we know what's good, we can recover the correct sequences. But that's, that's not the end of it, because these are just small little pieces of DNA. For this to be valuable They need to be longer, and so we need to assemble them, right? So we can do that multiple ways. We can inject beads into, like multiple beads into the same well and assemble them all in the same well, or we could inject them into different wells and then use robots to assemble them. But this is a pretty dirty mixture, so we have to assemble and then clean it up a bit. So,
